 |
New Questions, More Complex
Systems
GE Healthcare’s Nick Thomas
sees three trends pushing HCS earlier. “The assays are becoming more robust
and dependable, more groups have made a big investment and want to get more out
of their equipment, and finally, people are looking at nontraditional
targets.”
|
|
PROS AND CONS OF PARALLEL
SCREENING. Merck research fellow
Alexander Szewczak. |
Many types of labs are tackling
much more challenging problems these days. Academics are finding they finally
have the tools to look at more interesting biological questions. Pharmaceutical
companies, meanwhile, are finding that to reach fruitful targets they must look
further afield, into complex uncharted territory.
Scientists are also
increasingly looking at “more physiologically relevant cells, such as primary
cells today and, in the near future, stem cells,” Thomas says. That means the
cells under study are both more heterogeneous and their readouts may be less
obvious. While traditional high-throughput screening provides a single readout
per cell, with HCS, multiple signals can be obtained.
As a result, while much focus
is directed to HCS instrumentation, Thomas feels that “some of the most
important recent advances are in software and reagents.” GE Healthcare and
other players in this field are doing extensive product development in these
areas.
For example, getting some kind
of readout from HCS is predicated on getting cellular reporters or sensors —
such as fluorescent antibodies — into cells. But it is harder to get such
molecules into the primary cells increasingly used for these studies. With this
in mind, GE Healthcare developed its new line of Ad-A-Gene vectors. The vectors
make a tedious and iffy process simpler and more productive. “We’ve done all
the molecular biology,” Thomas says. “You take the virus product from the
freezer, add it to your cells, insert the test compound, and image.” The line
includes two products — some tagged with green fluorescent protein, and
another set carrying a response element. The first gives a green readout, the
other red. “If you use both you can see both the activation of the signaling
pathway and the arrival of the protein in the nucleus,” he explains.
Keeping up with the trend
toward collecting more and different types of information, Cellomics recently
launched its VTI Live module for the ArrayScan VTI HCS reader. This new module
lets researchers run automated live cell and kinetic assays and get time-course
readouts. “With the new software, scientists can now capture fast or
long-acting biology,” explains Kevin Gutshall, HCS instrumentation senior
product manager at Cellomics, a business unit of Fisher Scientific.
The instrument features a
special chamber, with temperature control, CO2, and humidity so that
cells are maintained in an incubator-like environment. Breakthrough kinetics
software analyzes images and data across a time course determined by the
experimenter. “Temperature, humidity, and CO2 need to be strictly
controlled to keep the cells healthy — the VTI Live module allows for precise
control of all three of these items,” says Gutshall. “If the instrument is
hooked up with Thermo Electron’s HCS Workcell, researchers can take the plates
in and out without putting them through any harsh exposure.”
Lang also emphasizes assays and
informatics. “High-content screening is reliable assuming you know what you
are measuring,” he says. “Behind every image, there are 10 to 50 parameters
to choose from, and you have to remember they are linked to a biological
process.” Many users are looking at only a couple of parameters. To use many
more requires top-shelf tools, according to Lang, particularly bioinformatics.
His group has been using data analysis and visualization tools from Gene Data.
“There was definitely a learning curve,” he says. “But we got real value
from this.”
At MRL, the rapidly growing
Lead Optimization department is also working around the challenges posed by the
new approach. “Although parallel screening can get you a lot of important
information, you have to remember it is based on cell-based assays, which are
not perfect models,” says Szewczak. “They are good models, but they are not
perfect for what happens within living organisms.” Researchers have to do
complementary low-throughput studies, he advises, to help round out the data.
That can slow things down, but
new robots are helping out. The MRL group has a workstation approach, with
readers, liquid handlers, GE IN Cell Analyzer 1000 HCS instrumentation, and
automated compound handling. “The Automation Partnership’s SelecT system,
which is an automated cell-culture instrument, has proven very useful for us,”
notes Szewczak. “It’s nice for obtaining uniform samples.” Because the
instrument runs 24/7, “You can cut down on the amount of extra time anyone has
to put in, such as work on weekends,” he adds.
Innovations
Beyond improvements in HCS,
better assays, and better robotics, researchers are also hoping for some
completely new approaches that could power up parallel screening.
One development that Lang and
other experts are watching closely is emergence of new label-free approaches.
“Label-free screening is a very clear future trend,” he says. Labeling can
interfere with cellular processes or readout, causing yet another complication
in an already tricky process. Several companies including Akubio, Biacore,
Corning Life Sciences, and MDS Sciex are developing such products.
“There are many label-free
approaches that work reasonably well,” explains Ron Verkleeren, a product
manager at Corning Life Sciences. “But nobody is doing that kind of screening
in large volume because of cost and time.”
Corning ’s just-launched Epic System not only works with ligands that are hard to
label, it’s also high throughput. The System avoids chemical labels by
exploiting properties of “glass and light,” Verkleeren explains. Binding
data are generated by measuring the light reflected back through a particular
type of glass. The system, which uses 384-well microplates, reads up to 40,000
wells in eight hours. Corning is developing protocols for biochemical, cellular, and hybrid-based assays.
Verkleeren says, “We see Epic as complementary to HCS. It has many
applications, including pathway analysis.”
With a label-free system such
as Epic, researchers can skip some steps, such as having to engineer cells to
over-express a receptor/target. “You hear horror stories about customers who
have done a lot of screening with over-expressed cells, only to discover the
compounds were only effective when the receptor is over-expressed,” Verkleeren
says.
Biotechnology company Cellzome
is offering a specialized new label-free screening approach built around its
proprietary Kinobeads. Cellzome uses the platform for its own discovery and with
collaborators. The proteomics-based system includes a number of broad-spectrum
kinase binding compounds that are immobilized on to beads. These beads are
incubated with ground-up tissue, and they affinity-capture hundreds of kinases
and related proteins from the sample. When compounds are added, they must
compete with the kinases to bind the beads. “We end up with something close to
a set of ‘proteomic’ affinity constants for all the targets,” says Gitte
Neubauer, vice president of research operations and a founder of Cellzome.
The system can test binding of
a drug or lead compound against 200 different kinases in a single tissue sample.
Given how many companies are wrestling to determine kinase selectivity, this
could prove to be a valuable approach. “With our system there is no
modification of the compound or the target protein,” Neubauer adds.
The system has other advantages
as well, according to Gerard Drewes, director of discovery research at Cellzome.
It can be used to screen chemical libraries even when there is no recombinant
protein against the target available; it provides compound binding data against
the target’s activated state; and the same assay can be used in cellular
assays for lead optimization or in animal studies. “Nobody else is screening
kinases in tissues,” says Drewes. “Everyone else has to use purified
recombinant proteins or even artificial protein fragments.”
The National Institutes of
Health’s Chemical Genomics Center (NCGC), meanwhile, has introduced a new type
of high-throughput screening that may itself provide more, and more valuable,
information. Quantitative high-throughput screening, or qHTS, is a simple twist
on the old paradigm. Instead of dropping compounds in at a single uniform dose,
the NCGC researchers tried using seven different doses. The result was a
remarkable drop — 35 percent to 75 percent fewer — in the number of false
negatives, or compounds that would have been incorrectly deemed unusable.
“Of course, it takes longer
to screen compounds at multiple doses,” says Chris Austin, NCGC director. But
the drop in false positives is so dramatic that the process is still more
efficient, he argues. The group has published a paper on the new approach, which
they run on a Kalypsys ultra-high-throughput screening system using 1,536-well
microplates.
A Top-Down Decision
Even with all these new tools
spread before them, pharmaceutical and biotech companies do not have an easy
choice. “It ends up being a resource allocation issue,” says Lang. Because
parallel screening costs are more up front, “The decision has to come from the
top,” he says. For a biotechnology company such as Serono, it is probably
easier to argue that it is critical to understand biology of the drug/target
interaction as soon as possible. Large pharmas, which are traditionally
chemistry driven at this stage, may have a harder time making that point. Most
of those companies have instead been trying to improve their success rates by
using bigger and bigger screening decks generated by combinatorial chemistry or
parallel synthesis.
Still, those bigger screens
have not been doing the trick. And groups using parallel screening with smaller
decks, and a more iterative approach, say they are not just getting better drug
leads, they are also netting invaluable structure/function data early on, which
is something chemists have long been clamoring for. (See “Pfizer’s Global
Survey of Pharmacological Space,” September 2006 Bio•IT World.) That
kind of information could give researchers the confidence they need to abandon
the pure “compound numbers” game and try the parallel approach using
“intelligently” selected compound decks. Lang, for one, has made his bet.
“We are all driven by how to improve our attrition rates now,” he says.
“We think parallel screening is one good tool to do that.”
Laurie Sullivan, senior
technology editor, Pharma DD, contributed to this report.
|
Sharon Terry
|
Nonprofits Share Pharma’s
HTS Woes
The cost, complexity, and
uncertainty of high-throughput screening is a major problem for pharmaceutical
companies, but for patient groups it’s an agony. Take Sharon and Patrick
Terry, founders of PXE International.
|
Since 1995, when the Terry’s
first learned their two children had inherited PXE (pseudoxanthoma elasticum),
the couple has basically reengineered the “activist parent” paradigm,
rolling up their sleeves and taking part in almost every aspect of the search
for a PXE cure. They founded PXE International, helped clone the gene involved
and patented it, and are giving financial and intellectual support to about 30
scientists doing PXE-related research. As laymen in a science-driven field,
they’ve made extraordinary progress, and it would seem that they have opened
all the necessary doors to find at least one prospective treatment. But now, as
Patrick Terry puts it, “We’re up against a formidable wall.”
That wall is all too familiar
to pharmaceutical companies: It is the problem of getting a good assay for a
nontraditional target and then getting your hands on a reasonable set of
“drug-like” compounds to test against it. “As long as you’re getting
assays off the shelf, it’s fine,” says Serono’s Paul Lang. “Developing
them yourself can be extremely difficult, and is sometimes impossible.”
Few off-the-shelf assays exist
for rare diseases, which often involve poorly studied molecular targets. To make
things worse, academics have only recently had even limited access to truly
high-throughput screening equipment. As a result, few have any training in
designing robust high-throughput screening assays, while still getting
reproducible results over hundreds of thousands of data points.
To be at all useful, assays
need to measure exactly the right process, be easy to read, and be reproducible.
Just validating them is a costly and tedious process. Serono, for example,
determines six key characteristics for each assay, including optimal cell growth
conditions, incubation times, and passage number. Each of these features is
tested and retested under multiple conditions at different times and with
different researchers using them.
To keep the effort to cure PXE
moving on multiple fronts, Sharon Terry, who also heads the nonprofit Genetic
Alliance, is trying to encourage pharmaceutical companies to participate in what
she calls a “compound/target dating game.” Nonprofit groups would provide
information about targets they’ve investigated, and companies could match
those targets to compounds that have been sidelined for lack of a market
opportunity. The idea is not unique. CancerResearch UK
is trying a similar approach in England and says it is hoping to get several partnerships per year around such
“shelved” compounds.
Sharon Terry has been in
discussion with Pharmaceutical Research and Manufacturers of America (PhRMA)
about the database for a couple of years. The plan, a PhRMA spokesman says, is
“still under consideration,” but no firm next steps are yet in place. It’s
easy to imagine the hurdles to this project. For one thing, drug companies are
concerned compounds will be tarnished if they show any toxicity, even in a group
as unique as patients with rare genetic diseases. “That’s a problem Congress
will have to tackle for us,” says Sharon Terry.
She is now pitching her
database idea to other foundations that may be able to move more quickly than
PhRMA. The Terry’s and other groups like PXE International are also trying to
help improve the quality of the assays academics work with.
Another group working
specifically on that problem is NIH’s Chemical Genomics Center, which offers information and guidance on assay development. “Rare
diseases fit perfectly with our mission,” explains Chris Austin, director of
the Center. “Not only is it right to work on these from a humanitarian
perspective, but understanding these single-gene disorders will give us profound
biological insights and will likely point to treatments for common diseases as
well.” As growing numbers of academics push to make their basic research
findings medically relevant, more of them may also bump up against this hurdle.
Perhaps that will bring new urgency, and even more new approaches, to the search
for simpler, quicker, and better compound screening and selection methods. M.A.B.
|
Selected References
The EPIC System:
Fang, Y. et al. “Characteristics of
dynamic mass redistribution of epidermal growth factor receptor signaling in
living cells measured with label-free optical biosensors.” Anal Chem
77, 5720-5; 2005.
Fan, Y.l. et al. “Optical biosensor
provides insights for bradykinin B2 receptor signaling in A431 cells.” FEBS
Lett 579, 6365-74; 2005.
NCGC’s qHTS:
Inglese, J. et al. “Quantitative
high-throughput screening: a titration-based approach that efficiently
identifies biological activities in large chemical libraries.” Proc Natl
Acad Sci USA 103, 1147-78; 2006.
HCS for Toxicity Screening:
O’Brien, P. et al. “High
concordance of drug-induced human hepatotoxicity with in vitro
cytotoxicity measured in a novel cell-based model using high content
screening.” Arch Toxicol (to be published April 6, 2007).
|
